Genomic DNA has to be replicated before every cycle of cell division. Although replicative DNA polymerase has a built-in proofreading mechanism to minimize the errors during replication, occasionally mismatch due to replication-errors still happens. Mismatch repair systems to prevent such mutations from occurring exist in most organisms. E. coli has a methyl-directed mismatch repair system comprising MutS, MutL and MutH proteins. Homologues of MutS and MutL proteins are also found in human. Mutations in these proteins are identified in 90% of the hereditary nonpolyposis colorectal cancers. In the three years before Oct. 2000, our group determined the crystal structures of mismatch repair proteins MutH, a conserved 40KD ATPase fragment of MutL and its complexes with nucleotides, and the 190 Kd Taq MutS alone, complexed with DNA, and as a ternary complex with DNA and ADP/Mg2+. Last year (2000-2001), (1) we carried out mutagenesis and biochemical studies of E. coli MutS and based on our results proposed an ATP-assisted mismatch recognition proofreading mechanism, which is similar to the proofreading in protein synthesis, (2) we determined the crystal structure of the ATPase fragment of Pms2, a human homolog of MutL and characterized its unusual monomeric ATPase activity, (3) in collaboration with Dr. Thomas Kunkel, we initiated the studies of asymmetric DNA binding and ATPase hydrolysis in heterodimeric eukaryotic MutS homologs and characterized the residues essential for mismatch recognition versus residues required for general protein-DNA interactions, (4) we generated 47 functional mutations in E. coli MutS, MutL and MutH based on our crystal structures and in collaboration with Dr. Jeffrey Miller of UCLA, we studied their in vivo mismatch repair and DNA recombination phenotypes. Combined with our in vitro studies, we have produced a comprehensive profile of the function and mechanism of these mismatch repair proteins. A new family of DNA polymerases, the Y-family, has recently been identified. They differ from the previously known DNA polymerase in the primary sequence and in lesion-bypass and error-prone DNA synthesis. In collaboration with Dr. Roger Woodgate of NICHD, we have determined the crystal structure of a Y-family DNA polymerase, Dpo4 from S. Solfotaricus, in complex with the substrate DNA and an incoming nucleotide. The crystal structures provide the first view of the Y-family polymerase in action and reveal the molecular mechanism for low fidelity DNA synthesis and bypassing modified DNA bases. We have continued our research of V(D)J recombination. V(D)J gene rearrangement in vertebrates is essential for the maturation of immune systems, which allows the generation of antibodies and T-cell receptors to build up the defense system. V(D)J gene rearrangement is a type of site-specific DNA recombination. Two proteins, RAG-1 and RAG-2 (recombination activation gene products), are necessary and sufficient to turn on the gene rearrangement in vivo, but they are difficult to obtain in a pure and soluble form. We have cloned and produced several soluble fragments of the mouse RAG1 proteins. Even though soluble, pure and in ample supply, these fragments still resist our crystallization attempts. Interestingly, one of the fragments contains a nuclease activity, which might have cellular functions. In collaboration with Dr. Marjorie Oettinger of Massachusetts General Hospital, we are characterizing the nuclease activity.